Setup of a new quad copter, two issues: high vibe values at high speed and unstable vertical descent

I have built a long range liion drone (AUW 2.8 kg, battery 6S6P 20 Ah, 380 Kv sunnysky 3508, 13 x 5). it flies quite well, has an endurance of 80 minutes/45 km.


It runs on ardupilot.

Now, I have run into a problem that with an increase of speed, the vibe values increase.

On hover or slow speed, Y vibe and Z vibe are low (5) and X vibe around 10.
On medium speed 50 km-h, Y goes up to 10, and Z and X go to around 30.
On high speed 80…110 km-h, Y goes to 20, X goes to 40, and Z to 80 and even higher.

On one occassion, I have lost control, the AP gave something like a vibe reset, the whole thing seemed to have lost control, but then somehow recovered.

This behaviour seems not to be related to the power of the motors, but rather horizontal speed. On the occasion where I lost control, I was very high, and decided to dive down in acro mode, GPS speed accelerated to about 110 km-h, but the motor current was around 12 A, which is about the same as needed for hover.

I am not sure where the problem could be. The arms are not particularly strong (front arms are 16 mm diameter, 1 mm wall thickness, and rear armas are 12 mm diameter, 3 mm thickness), and they are somewhat flexible (you can twist the arm if you grab the motor), but at low speeds, and even high power (like in a climb at 6 m/s) the vibe values are low. So I am not sure if this is the cause of the problem

Today I modifoed the FC suspension, adding a 1 cm. foam padding. Then i flew around accelerating and decelerating. The result of the padding seems to have reduced the vibes by about 50 %, what was going up to about 50 60 at a given speed now is around 30 … 35.

Still, I have rather high vibe numbers when flying at higher speed.

Also, when flying forward, the X vibe is much stronger then when flying backwards at the same speed (speed cycles on the graph 1 and 3 are forward, cycle 2 and 4 are backwards).
When flying sidewise the X vibe is average.

I am inclined to consider stregthening the rear arms, but still not sure. I do not like the idea, because it will introduce more weight and will reduce the endurance. The whole quad weighs exactly 1000 grams empty, of which the motors with props take up 500 grams.

I am not that bothered by the flight behaviour in the normal speed range, i.e. like flying up to 70 km/h or 20 m/s seems to function perfectly.

I am worried that if I have to go beyond that envelope, I may enter unrecoverable situation. That is especially true for rapid descents, since it takes ages to hover down (I feel like max vertical speed should not be higher than 2.5 meters), and the way to go is to slide down in acro mode, where I can easily reach 10…12 m/s vertical speed, but that requires a very precise throttle control so as not to go too fast horizontally,

Second issue I have is very unstable vertical descent in Loiter and Alt Hold modes. At vertical speeds of even 2 m/s it oscillates strongly, but without getting totally out of control. Alt Hold seems to be slightly more stable than Loiter.

This is the graph of roll for descent in Loiter and Alrt hold:

And this is a graph of roll during normal horizontal flying:

I have Autotuned the copter, I have slightly reduced PI for pitch and roll by about 20 % and in general flight it feels pretty good.

I have also tried vertical descents in acro mode at speeds of up to 7 m/s, it shakes but seems to be more stable than in Loiter mode at 2 m/s vertical descent.

I understand that when descending through prop wash, you should expect some instability, but in this case only at speeds of about 1.2 meters the descent stabilizes, so I suspect that one has possibly to tweak the PIDs values.

For the time being to kill off altitude I fly in acro mode and inclined by about 30 degrees with power setting close to normal hover, this allows descent rates of 8…12 m/s with quite stable behaviour.

I added some 1mm carbon rod to stop my frame twisting and bending under load.
28e0ff05804a29189ca0f8e863e15fa947716938_2_666x500

Did it help? What parameters have this changed? Vibe?

yes, the arms were getting twisted by the torque of the motors, you can lower the imu filter to tune it out but its not ideal

But how exactly did it improve the thing, did have oscillations which went away?

In my case, i can twist by hand the arms, but I am not sure that motors can do that. It seems that I have wind buffetting causing vibration, not motors. (The vibe values are indpendent of power setting of the motors)

it’s reverberating, the shaking is building up because its trying to correct it and that makes it worse because it ends up in sync with the shaking causing it to exacerbate the problem by bending it further the other way.

I feel that this is not my problem…

if your post you literally say this:

they are somewhat flexible (you can twist the arm if you grab the motor)

but you dont think that could be related to the problem??

Two reason why not. First, I have to very strongly twist it by hand to turn the outer portion by maybe 2…3 degrees. Second, the problem I have is not power related, but speed related. I.e. if I apply maximum thrust or if I do violent forward-backward stick movements, that does not affect the vibes. But high speed does, and the higher it goes, the worse it gets. I am inclined to do a run in a car with open window and stick the copter out to see how it would behave with motors off.

I don’t know what country you’re in but most countries do not allow long range UAV flights.

Did not have a clue about that… I thought DJIs can fly about 8 km…

They can, but not legally, at least not here in the USA.